On Wed, Aug 27, 2008 at 3:50 AM, Martin Thomas <[EMAIL PROTECTED]> wrote:
> Hallo,
>
> I have just played a little bit with a LPC2378 on an Olimex LPC-2378-STK
> board using OpenOCD from SVN 971 (built with mingw/msys, running on
> WinXP with JTAGkey). The calc_checksum option is given in the flash bank
> definition and 'flash write_image erase project.elf elf' is used to
> flash the target followed by 'verify_image project.elf elf' to see if
> the flashing has been successful. OpenOCD has calculated the checksum
> correctly and the application works as expected but verify fails since
> the "on-the-fly" modification is not taken into account:
>
>> verify_image project.elf elf
> checksum mismatch - attempting binary compare
> Verify operation failed address 0x00000014. Was 0x58 instead of 0x00
>
> It would be nice the rest of the flash-content could be verified even if
> the checksum at 0x14 does not match the content of the load-image.

The # of mismatches could be VERY large, so obviously there
must be some limitation mechanism here.

I'm not sure I know what would make the most sense here...

The current best practice would be to download the data and
diff them on the dev pc...

> As a
> quick workaround maybe like this
> - check if an LPC2k "bank" is used
> - check if the calc_checksum option is given
> If both "true": on mismatch at 0x14 just print "Warning: LPC2000
> check_sum at 0x14 mismatch ignored" and just continue verifying the rest
> of the sections


I'm not familiar with the calc_checksum option. Sounds strange to have
that built into the driver.

>
> As far as I can see gdb's compare-sections also alerts the mismatch.
>
> Martin Thomas
>
>
> _______________________________________________
> Openocd-development mailing list
> [email protected]
> https://lists.berlios.de/mailman/listinfo/openocd-development
>



-- 
Øyvind Harboe
http://www.zylin.com/zy1000.html
ARM7 ARM9 XScale Cortex
JTAG debugger and flash programmer
_______________________________________________
Openocd-development mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/openocd-development

Reply via email to